libcfg+ - Viewing Bug Advanced Details			[ Go Back ]
ID:	Category:	Severity:	Reproducibility:	Date Submitted:	Last Update:
199	library core	major	sometimes	2004-03-08 10:37	2004-03-26 14:41
Reporter:	admin	Platform:
Assigned To:	nepto	OS:
Priority:	normal	Version:
Status:	resolved	Product Version:	0.6.2 (stable)
Product Build:		Resolution:	fixed
Projection:	none	Duplicate ID:
ETA:	none
Summary:	last line in config file is not parsed
Description:	thanks for the ecellent library, by far the best features and best usabilty for option and command line parsing I have seen so far. There is a little bug in config file parsing, apparently a missing newline on last line screws up: {NULL, 0, "threading", CFG_ULONG, (void *) &threading, 0}, ... ... con = cfg_get_context(acct_options); cfg_set_cfgfile_context(con, 0, -1, cfg_file); ret = cfg_parse(con); if (ret != CFG_OK) { printf("error parsing config file %s: ",acct); cfg_fprint_error(con, stdout); putchar('\n'); exit(1); } Now when "threading=1" is the last line in config file (without newline) it doesn't set "threading" and it doesnt complain either. After adding a newline in the config file it works as expected.
Steps To Reproduce:
Additional Information / Solution:	This bug was reported via e-mail by Richard Zidlicky <rz AT linux-m68k.org>
Attached Files:	0000199-threading.tgz.txt (2 KB) 1970-01-01 01:33 0000199-libcfgbug.tar.bz2.txt (1 KB) 1970-01-01 01:33 0000199-big-199-patch.diff.txt (1 KB) 1970-01-01 01:33

Bug Notes
nepto    2004-03-15 11:33	The problem is, that bug cannot be reproduced. I created tiny testing program, "threading" variable is still successfully parsed, even if there is no newline at the end of file. 1. Download threading.tgz file. 2. Unpack it in the libcfg+ root directory. 3. It will create threading/ subdirectory for you. 4. Go in and type "make". 5. Than run ./threading binary.
nepto    2004-03-17 14:53	Richard's reply: On Mon, Mar 15, 2004 at 11:40:15AM +0100, Ondrej Jombik wrote: Hi, thanks for looking at it. > Probably you are using some special libcfg+ context flags or some older > version? nothing fancy, I use version 0.6.2. Tested your testprogram and it says "threading = 1". However a closer look at "threading.cfg" revealed that it *has* a newline at the end of file.. perhaps some helpfull editor added it. # od -c threading.cfg 0000000 # \n # B u g # 0 0 0 0 1 9 9 0000020 : l a s t l i n e i n c 0000040 o n f i g f i l e i s n o 0000060 t p a r s e d \n # h t t p : 0000100 / / p l a t o n . s k / p r o j 0000120 e c t s / b u g _ v i e w _ a d 0000140 v a n c e d _ p a g e . p h p ? 0000160 f _ b u g _ i d = 1 9 9 \n # \n \n 0000200 # n o t e n o n e w l i n 0000220 e a t t h e e n d o f 0000240 t h i s f i l e \n \n t h r e a 0000260 d i n g = 1 \n ^^^^^ After deleting that newline I get # ./threading threading.cfg configuration file was successfully parsed threading = 0 Attaching my testcase which is almost identical to yours anyway, wants libcfg+ installed in standard location and "make" will do everything including running the tests. Richard
nepto    2004-03-26 12:11	A patch fixing this issue was uploaded.
nepto    2004-03-26 14:41	Commited into CVS.